Research On State Of Charge Estimation Method Of Lithium-ion Battery Using Wavelet Neural Network

The world is facing two major problems which are increasingly serious: energy shortages and environmental degradation.Finding a greener,more efficient and low-carbon energy based transportation program has become a hot research filed.As an important part of transportation,the automotive industry has ushered in a new period of development and transformation.The development of new energy vehicles has become a historically inevitable choice.As the main energy carrier of electric vehicles,the power batteries have undoubtedly become the research and development frontier of new energy vehicles.Lithium-ion batteries have become the preferred energy carrier for electric vehicles due to their high energy density,high output power,good cycling performance and no memory effect.Nevertheless,during manufacturing or using process,the individual difference between the cells cannot be ignored.The cell difference directly affects the performance of the power battery pack and causes the safety problem.Therefore,it is of great significance to manage and control all battery cells.State of charge(SOC)is the core parameter of the battery management system,whose estimation is the top priority and has become a research hotspot in the new era.This paper takes lithium-ion batteries as the research object,tests the battery's multi-simulated driving cycles characteristic,and analyzes and validates the performance of the SOC estimation algorithms based on the proposed improved wavelet neural network model using a number of experiments.The main contents of this paper are listed as follows:1)Combining discrete transform with adaptive wavelet neural network,a hybrid adaptive wavelet neural network is proposed.Experiments show that the hybrid adaptive wavelet neural network based SOC estimation algorithm has excellent accuracy,robustness and stability.2)A multi-hidden-layer adaptive wavelet neural network is proposed to increase the learning depth of network and applied to the SOC estimation process of lithium-ion batteries.The performance of multi-hidden-layer adaptive wavelet neural network based SOC estimation method is analyzed and studied through experiments.Through analyzing the experimental results,a general improvement method for adaptive wavelet neural network based SOC estimation method is proposed.3)The particle swarm optimization algorithm and Levenberg-Marquardt algorithm are employed as the learning algorithm of the proposed wavelet neural networks.Through the application of SOC estimation,the practical application effect of the learning algorithm is compared and analyzed.4)Based on the hybrid multi-hidden-layer adaptive neural network,preliminary analysis and experiments are conducted for a variable temperature working condition.It is proved to some extent that the neural network based SOC estimation methods have the advantages of simple modeling and high applicability.With the wide application of artificial intelligence technology in kinds of fields,intellectualization has become an inevitable trend of the times.Artificial neural network based intelligent SOC estimation methods will also become the cornerstone of the intelligent battery management system.Therefore,the series of the improved wavelet neural network based intelligent SOC estimation methods proposed in this paper is of great significance.